The problem of predicting plume rise (and subsequently ground level concentrations) from a single source has been illustrated here by a comparison of the integral plume rise and dispersion model (U.S.P.R.) with several data sources. One advantage of this formulation is that it allows the simulation to incorporate the effects of continuously variable stratification, including inversion layers, meteorological characteristics etc. which empirical formulae for plume rise in general cannot include in any realistic fashion. Furthermore, it permits a simulation from source to sink without recourse to any empirical concatenation of separate plume rise and dispersion models.